search

mikeblazanin / gcplyr

gcplyr is an R package that facilitates wrangling and analysis of microbial growth curve data
https://mikeblazanin.github.io/gcplyr/
Other
30 stars 2 forks source link

additional discussion of smoothing considerations in vignette #68

Open mikeblazanin opened 1 year ago

mikeblazanin commented 1 year ago

Discussing the statistical background of smoothing, what data should be included on Dryad, how to run stats on smoothed data, what data should you show your PI?

A preface about data manipulation concerns

Smoothing is necessary for calculating derivatives basically

Does smoothing p-hack?

mikeblazanin commented 1 year ago

Alita: "lengthy discussions about data integrity, transparency, and statistics" needed

mikeblazanin commented 1 year ago

In particular, statements like "some degree of smoothing before analysis is necessary to reduce this noise and improve the accuracy of analyses" draw Alita's reply that "This sounds like data manipulation to me. Stats should take care of this through analysis of many replicates"

mikeblazanin commented 1 year ago

This issue is much improved in the latest version of the vignette. However, discussion of how this is necessary to resolve signal out of the noise is probably still lacking. Would be nice to find citations to support that argument